Technical Field
[0001] The invention relates to a method and an apparatus for determining a mobility of
ions. The method includes the steps of modulating an ion beam with an ion gate which
is controlled by a modulation function for generating a modulated ion beam, of guiding
the modulated ion beam through a drifting region, of measuring a signal of the modulated
ion beam after the modulated ion beam has passed the drifting region and of calculating
a correlation of the modulation function and the signal in order to determine the
mobility of the ions. The apparatus includes the ion gate, the drifting region through
which the modulated ion beam is guidable, a detector by which the signal of the modulated
ion beam is measurable after the modulated ion beam has passed the drifting region
and a calculation unit by which the correlation of the modulation function and the
signal is calculable in order to determine the mobility of the ions.
Background Art
[0002] Methods and apparatuses pertaining to the above mentioned technical field are known.
For example, in
US 2009/0294647 A1 (Karsten Michelmann), an ion mobility spectrometer which is coupled to a mass spectrometer and a corresponding
measuring method are described. The ion beam is modulated with a continuous modulation
function and the modulation frequency of this modulation function is varied over a
large frequency range. In order to obtain the ion mass spectrum, the measured ion
spectrum is correlated with the modulation function of the ion beam.
[0003] Another ion mobility spectrometer and a corresponding method are disclosed in
US 7,417,222 B1 (Sandia Corp). There as well, the ion beam is modulated with a modulation function
and the measured signal is correlated with the modulation function. But in contrast
to
US 2009/0294647 A1, the modulation function may also be a binary function. In particular, Barker codes
are described as being favourable modulation functions because their autocorrelation
provides low side bands.
[0004] A somewhat different approach is described in
US 6,900,431 B1 (Predicant Biosciences, Inc.) on the example of a time of flight mass spectrometer.
Here, the ion beam is modulated in pseudo random sequences of maximum length. The
characterisation of the ion spectra is obtained by the inverse Hadamard transformation
formalism. Similarly, in the ion mobility spectrometer disclosed in
WO 2004/097394 A1 (Smiths Group Plc), the ion beam is modulated with a pseudo random sequence of maximum
length and the measured ion signal is analysed by a matrix algebra. But in the latter
example, a Fourier analysis may be used instead of the matrix algebra, too. Additionally,
two modulation sequences with inverted bits may be used in order to obtain a better
signal to noise ratio.
[0005] These known methods have in common that the ion beam is modulated according to a
modulation function, that the ion signal is measured after the ions have passed a
drifting region and that the ion mobility is obtained by calculating a correlation
of the modulation function with the measured ion signal. This procedure for obtaining
the ion mobility is employed because it is not required to know the starting time
of each individual ion as it would be if directly measuring the ion's flight time.
Consequently, it is possible to pass at the same time more than one pulse or packet
of ions through the drifting region. This has the advantage that more ions can be
measured within the same period of time.
[0006] The disadvantage of this procedure is that calculating the correlation introduces
features into the ion mobility spectra which cannot easily be identified as such.
For example, these features may be small peaks in the ion mobility spectra that look
like a signal obtained from some specific ion species. Therefore, if traces of ions
are to be detected, such artificially introduced features are likely to lead to misinterpretations
of the ion mobility spectra. Thus, in order to avoid such misinterpretations, small
peaks in the ion mobility spectra have to be discarded as possible false peaks. This
significantly limits the attainable dynamic range.
Summary of the invention
[0007] It is the object of the invention to create a method and an apparatus pertaining
to the technical field initially mentioned that allow for determining an ion mobility
with a higher signal to noise ratio while providing the same measurement speed as
known from the prior art.
[0008] The solution of the invention is specified by the features of the independent claims.
According to the invention, the autocorrelation of the modulation function is a two-valued
function. This means that the autocorrelation function has a peak at zero and a constant
value at all other points.
[0009] The advantage of the modulation function having a two-valued autocorrelation function
is that calculating the correlation does not introduce additional features into the
ion mobility spectra.
[0010] Preferably, the modulation function is a binary function. Accordingly, the modulation
function may be represented by a row of bits. This has the advantage that it is simple
to modulate with the ion gate the ion beam such that in flight direction of the ions
the modulated ion beam has the shape of the modulation function. In a variant, the
modulation function is based on a binary function but provides smoothed steps between
the bits of the binary function. This has the advantage that depletion of ions in
a region behind the ion gate and tailing or diffusion of ions in the modulated beam
can be taken into account for by adapting the modulation function to these effects
in the modulated ion beam before calculating the correlation. In a further variant,
the modulation function is based on a binary function but is oversampled. That is,
multiple measurements are made during each "0" and "1" of the binary function. Alternatively,
the modulation function is a non-binary function, which may also be oversampled.
[0011] In the following, there are passages where the modulation function is described as
being a binary function or a sequence. In these passages, the modulation function
may effectively be the described binary function or sequence. But it may as well be
a function which is based on the described binary function or sequence. In the latter
case, the modulation function may provide smoothed steps between the bits of the described
binary function or sequence and/or may be oversampled.
[0012] Preferably, the modulation function is a pseudorandom sequence. This has the advantage
that the properties of the modulation function approximate the properties of a random
sequence. Therefore, repetitions in the modulation function that would lead to additional
peaks in the ion mobility spectra can be avoided if the length of the pseudorandom
sequence is chosen accordingly. Furthermore, a pseudorandom sequence as a modulation
function has the advantage that the modulation function can easily be generated like
for example with a linear feedback shift register.
[0013] If the modulation function is a pseudorandom sequence of the type known as maximum
length sequences or of a type that can be represented by one ore more maximum length
sequences, it is advantageous to use a linear feedback shift register for generating
the modulation function. In such a linear feedback shift register a number of feedback
patterns are possible, called tap sets of the linear feedback shift register. The
number of possible tap sets depends on the length of the particular linear feedback
shift register. The modulation function is generated with the linear feedback shift
register by choosing a tap set and a set of initial values. The set of initial values
is fed to the linear feedback shift register. Based on the set of initial values,
the modulation function is then generated by the linear feedback shift register according
to the tap set. Therefore, the modulation function depends on the tap set and on the
set of initial values.
[0014] As a variant, the modulation function may be generated in a different way. For example,
one or more known pseudorandom sequences or other modulation functions may be stored
in a data store. For each measurement, a particular modulation function stored in
the data store may be used.
[0015] In a further variant, the modulation function may be a different function than a
pseudorandom sequence. For example, it may be a random sequence. This has the advantage
that the function has the corresponding properties. Alternatively, the modulation
function may be a non-random function.
[0016] If the modulation function is a pseudorandom sequence, it is advantageously a maximum
length sequence, a GMW sequence, a Welch-Gong transformation sequence, a Quadratic
residue sequence, a Sextic residue sequence, a Twin prime sequence, a Kasami power
function sequence, a Hyperoval sequence or a sequence derived from 3 or 5 maximum
length sequences. This has the advantage that the modulation function is a sequence
with well known properties. In case the sequence is derived from 3 to 5 maximum length
sequences, it may for example be obtained by adding up the content of corresponding
bits of the 3 or 5 maximum length sequences. In that case, the addition of two 1s
or of two 0s may result in a 0, while the addition of a 0 and a 1 or of a 1 and a
0 may result in a 1 (bitwise NAND operation).
[0017] As a variant, the modulation function may be a pseudorandom sequence which does not
belong to one of these classes.
[0018] Preferably, if the modulation function is a binary function or a sequence, it has
a length of more than 15 bits, preferably more than 50 bits, in particular more than
100 bits. This has the advantage that the modulation function is long enough to enable
measurements where sufficient ions are being measured for obtaining meaningful ion
mobility spectra. Alternatively, the modulation function may have a length of 15 bits
or less. This may be advantageous if the time of a measurement should be short and
if there are sufficient ions available for obtaining meaningful ion mobility spectra.
[0019] Advantageously, the method comprises a step of enhancing the edges of the signal
with a filter by filtering the signal before calculating the correlation. This has
the advantage that the resolution of the obtained ion mobility spectra is improved
in that the correlation is sharpened.
[0020] Alternatively, the method may not comprise a step of enhancing the edges of the signal
with a filter before calculating the correlation. If the obtained ion mobility spectra
should be as close as possible to the effectively measured signal, leaving out the
step of enhancing the edges of the signal may be advantageous because the required
filtering is a treatment of the measured signal.
[0021] If the method comprises the step of enhancing the edges of the signal with a filter,
the filter is preferably an n-element finite difference filter, an edge enhancement
filter or a filter using a different type of sharpening algorithm. This has the advantage
that an enhancing of the edges of the signal is obtained with a known sharpening algorithm
which can be adjusted to the particular characteristics of the signal to be treated.
[0022] For example, in case the filter is an n-element finite difference filter and the
signal is measured in bins having a specific width in time, the filter may comprise
an algorithm having the form
where n is a measure for the width of the filter,
Di is the size of the signal's /
th bin and F
i is the filter-value's /
th bin. In order to obtain the filtered signal, each filter-value F
i is added to the corresponding bin D
i of the measured signal. When doing so, it is possible to multiply the filter-values
F
i and/or the signal D
i with a weight factor before adding the filter-values to the signal. For example,
such a weight factor may be based on n, the width of the filter, with 0 <= n<= n
max:
[0023] Of course, it is possible to use weight factors that are independent of the width
of the filter as well. Furthermore, it is possible to flatten the signal D
i before calculating the filter-value by convoluting the signal with a Gaussian or
any other smoothing function. This may be advantageous because otherwise, noise in
the signal may lead to errors in the filter-value.
[0024] If the signal is not measured in bins having a specific width in time but by storing
for each measured ion (i.e. for each event) the time passed since a starting time,
the signal may be rasterised to bins of a specific width in time before applying the
filter. Alternatively, if for each event the time is stored which has passed since
the starting time, the filter's algorithm may be adapted to take into account for
the time differences between the individual events instead of assuming bins having
a specific width in time. The parameter
n of the algorithm may then become a measure for the time interval within which events
are considered when calculating a particular filter-value
Fi.
[0025] In case the signal is measured or rasterised in bins having a specific width in time,
it is advantageous that
n, the number of bins considered, is adapted to the characteristics of the signal. If
the filter should be calculated rapidly, it may be advantageous to choose
n to be 1. In this case, the filter becomes a Laplace filter. Otherwise, if the signal
is neither measured in bins having a specific width in time nor rasterised accordingly,
it is advantageous to adapt to the characteristics of the signal the time interval
within which events are considered.
[0026] For example, in case the filter is an edge enhancement filter, it may comprise an
algorithm where a blurred signal is calculated by convoluting the signal with a Gaussian,
and where the difference between the signal and the blurred signal is added to the
signal. Similar to the method of unsharp masking known from digital image processing,
three parameters of the algorithm may be adapted according to the particular signal
to be treated. First, the width of the Gaussian may be adapted. Second, before adding
the difference to the signal, the difference may be multiplied by a weighting factor
that is adapted to the particular signal. Third, a threshold parameter may be defined
such that the filter is only applied if the parameter's value is above a certain threshold.
For example, the threshold parameter may be the deviation of the blurred signal from
the measured signal.
[0027] Advantageously, an interval of interest of possible ion drift times is chosen from
the correlation. This has the advantage that the interval of interest of the ion mobility
spectra may be displayed or used for further data treatment. Alternatively, no specific
interval of interest of possible ion drift times is chosen from the correlation. This
has the same effect as if the interval of interest is chosen to spread over the entire
correlation. Accordingly, this alternative has the advantage that all data may be
displayed or used for further data treatment, respectively.
[0028] If the correlation is calculated for an interval of interest of possible ion drift
times, the method preferably comprises a step of selecting the modulation function
such that as many as possible false peaks in the correlation are located outside of
the interval of interest. These false peaks belong to a group of features in the ion
mobility spectra that are already present in the measured signal in the form of imperfections
and/or noise in the signal. The imperfections may be caused for example by depletion
of ions in a region behind the ion gate, by tailing of ions in the modulated beam,
by diffusion of ions in the modulated beam and/or by inhomogenities or turbulences
in a gas flow in the drifting region. Such imperfections may lead to a change of the
shape of the modulated ion beam. Accordingly, they may lead to unintended features
in the measured signal. As a consequence of calculating the correlation, the feature's
positions in the ion mobility spectra may be shifted as compared to their positions
in the measured signal. The shifting behaviour depends on the feature and on the modulation
function. For example, if the modulation function is a sequence that is generated
by a linear feedback shift register, the positions of some features in the ion mobility
spectra are determined by the tap set of the linear feedback register while they are
independent of the set of initial values used for generating the sequence. In the
present context, the term "false peaks" is used for this particular group of features
in the ion mobility spectra. Consequently, it is advantageous to use a linear feedback
shift register for generating the modulation function and to use tap sets of the linear
feedback shift register where the positions of false peaks caused by specific features
are known. For example, tap sets may be preliminary evaluated for features which are
characteristic for the ion mobility spectrometer that is used for executing the method.
These characteristic features may be depletion of ions in a region behind the ion
gate, tailing of ions in the modulated ion beam, diffusion of ions in the modulated
ion beam and/or inhomogenities or turbulences in a gas flow in the drifting region.
Once the interval of interest of possible ion drift times is known, the tap set which
is used can be chosen such that the false peaks in the ion mobility spectra are located
outside of the interval of interest. This has the advantage that the chances of a
misinterpretation of the obtained ion mobility spectra are reduced.
[0029] Alternatively, it is possible to leave out the step of selecting the modulation function
such that false peaks in the correlation are located outside of the interval of interest.
This may be advantageous if the interval of interest is large and if the available
modulation functions would be too strongly limited by such a selection or if there
would be no corresponding modulation function available at all.
[0030] Preferably, the method comprises a step of selecting the modulation function such
that false features in the correlation have a low height. Similar to the expression
"false peaks", the expression "false features" is used in the present context for
a particular group of features in the ion mobility spectra that are already present
in the measured signal in the form of imperfections and/or noise in the signal. If
the modulation function is a sequence that is generated by a linear feedback shift
register, the position of a false feature in the ion mobility spectra depends on the
tap set of the linear feedback shift register and on the set of initial values used
for generating the sequence. In addition, the height of the false features depends
on the set of initial values used for generating the sequence.
[0031] Accordingly, it is preferable to choose the modulation function such that characteristic
imperfections like depletion of ions in a region behind the ion gate, tailing of ions
in the modulated beam, diffusion of ions in the modulated beam and/or inhomogenities
or turbulences in a gas flow in the drifting region result in a minimal height of
the false features in the ion mobility spectra. This has the advantage that the chances
of a misinterpretation of the obtained ion mobility spectra are reduced.
[0032] Alternatively, it is possible to leave out the step of selecting the modulation function
such that false features in the correlation have a low height.
[0033] Advantageously, the steps of the method are repeated in cycles. During each cycle,
the ion beam is preferably modulated with the ion gate being controlled by a different
modulation function from a set of modulation functions for generating a different
modulated ion beam. Furthermore, the correlation which is calculated during each cycle
is advantageously added to a total correlation in order to obtain the mobility of
the ions. This has the advantage that by choosing a set of different modulation functions,
noise and systematic errors in the measured signal can me averaged out in the ion
mobility spectra.
[0034] As a variant, it is possible to repeat the steps of the method in cycles while the
ion gate is controlled by the same modulation function. This has the advantage that
the statistics of the signal and thus of the ion mobility spectra is improved.
[0035] Alternatively, the steps of the method may be executed once only. This has the advantage
that the measurement time is shorter.
[0036] If the steps of the method are repeated in cycles, it is advantageous to perform
a preliminary step before repeating the cycles. In this preliminary step, the set
of modulation functions is preferably selected such that for each modulation function,
the false features in the correlation are located at different positions of the correlation
and thus the false features are averaged out in the total correlation. For example,
if the modulation function is a pseudorandom sequence and the modulation function
is generated by a linear feedback shift register, a tap set of the linear feedback
shift register may be chosen such that a height of the false features is minimal.
Subsequently, this linear feedback shift register may be employed to generate different
pseudorandom sequences by feeding it with different sets of initial values. This has
the advantage that the obtained pseudorandom sequences cause false features originating
from the same imperfection in the signal to be located at different positions in the
correlation. Accordingly, the systematic imperfections causing false features in the
ion mobility spectra can be averaged out. Furthermore, this has the advantage that
if the correlation is calculated for an interval of interest, the tap set of the linear
feedback shift register may be chosen such that false peaks in the correlation are
located outside of the interval of interest. In that case, false peaks may be avoided
in the ion mobility spectra and at the same time false features may be averaged out.
[0037] In a variant, it is possible to perform the preliminary step only once for determining
one set or different sets of modulation functions. These sets of modulation function
may be stored and then be employed for different measurements.
[0038] Advantageously, the correlation is calculated by calculating a circular cross correlation,
an inverse Hadamard-transformation a Fourier transformation, a Laplace transformation
or an M-transformation. This has the advantage that the correlation is calculated
by a known formalism. Alternatively, a different formalism may be employed as well
for calculating the correlation.
[0039] Preferably, the apparatus for determining the mobility of the ions includes a linear
feedback shift register by which a pseudorandom sequence is generatable for the use
as modulation function. This has the advantage that pseudorandom sequences are easily
calculable. For example, this linear feedback shift register may be an electronic
circuit or may be based on computer software. In another example, it may be included
in the calculation unit.
[0040] As a variant, the apparatus may include a store for storing the modulation function.
This allows for storing pseudorandom sequences that were generated by the linear feedback
shift register in the store. This has the advantage that the measurement speed can
be improved if the modulation function is stored in the store prior to the measurement.
Additionally, the store has the advantage that it allows for storing predefined pseudorandom
sequences or other modulation functions. Accordingly, the apparatus may include a
store but no linear feedback shift register. In this latter case, the apparatus for
determining the mobility of the ions may comprise another unit for generating the
modulation function. For example, this unit may be a unit that generates predefined
modulation functions or a unit that generates random sequences as modulation functions.
In a variant, the apparatus may not comprise such a unit either.
[0041] Advantageously, before the correlation is calculable, a filter for enhancing the
edges of the signal is applicable by the calculation unit to the signal. As a variant,
the apparatus may include a separate filter unit by which a filter for enhancing the
edges the signal is applicable to the signal. Both variants have the advantage that
the resolution of the obtained ion mobility spectra is improved. Alternatively, it
is possible that there is no filter for enhancing the edges of the signal applicable
to the signal.
[0042] Preferably, the apparatus comprises a control unit by which a repetition in cycles
of steps is controllable, the steps including generating the modulated ion beam with
the ion gate, guiding the modulated ion beam through the drifting region, measuring
the signal with the detector and calculating the correlation of the modulation function
and the signal. Furthermore, the apparatus preferably comprises a summation unit by
which a total correlation is calculable in order to determine the mobility of the
ions, the total correlation being a sum of the correlations calculated during the
cycles. Thereby, it is possible that the summation unit is a separate unit or that
it is included in the calculation unit. In both cases, the control unit and the summation
unit have the advantage that noise and systematic errors in the measured signal can
be averaged out in the ion mobility spectra by controlling the ion gate with a different
modulation function of a set of different modulation functions during each repetition
of the steps.
[0043] As a variant, the apparatus may comprise the control unit and the summation unit,
but the ion gate may be controllable by the same modulation function throughout all
repetitions. This has the advantage that the statistics of the signal and thus of
the ion mobility spectra may be improved.
[0044] Alternatively, the apparatus may not comprise such a control unit or such a summation
unit.
[0045] Advantageously, the detector is a mass spectrometer. This has the advantage that
for the same ions being measured a mobility spectrum and a mass spectrum may be obtained.
Alternatively, if no ion mass spectrum is required, the detector may be a detector
which only detects ions and does not measure an ion mass spectrum. The latter case
has the advantage that the apparatus is simpler and can be constructed cheaper.
[0046] In case the detector is a mass spectrometer, the detector is preferably a time-of-flight
mass spectrometer. This is advantageous because a time-of-flight mass spectrometer
can optimally be combined with the ion mobility spectrometer because a time-of-flight
mass spectrometer allows for measuring a large range of ion masses with a high scan
rate. In a variant, the detector is a quadrupole mass spectrometer. This is advantageous
if a small range of ion masses is to be determined, where a high scan rate of the
quadrupole mass spectrometer may be obtained. In a further variant, the detector is
an ion trap mass spectrometer. Alternatively, the detector is a different type of
mass spectrometer.
[0047] In case the correlation function is a binary function or a sequence and the detector
is a mass spectrometer, the mass spectrometer preferably allows for determining ion
mass spectra with a repetition rate that corresponds to the bit length of the correlation
function. This has the advantage that the scan rate of the mass spectrometer is adapted
to the ion mobility spectrometer.
[0048] In an advantageous variant, if the detector is a mass spectrometer, the mass spectrometer
preferably allows for determining ion mass spectra with a repetition rate that corresponds
to a time resolution of the obtainable ion mobility spectra or to a fraction thereof.
This has the advantage that the scan rate of the mass spectrometer is optimally adapted
to the ion mobility spectrometer.
[0049] The above described invention may be employed as well in the fields of single and
tandem liquid and gas chromatography, when a time-of-flight mass spectrometer is used
as a detector. In these devices, the retention time of a substance on a column is
measured. This is conceptually and functionally equivalent to the ion drift time in
an ion mobility spectrometer. Usually, in order to obtain this measurement, the sample
is injected onto the column as at a known time, and elutes as a single peak after
the substance-specific retention time. When employing the invention in such devices,
the injection of the sample onto the column is modulated in time with a modulation
function that has an autocorrelation which is a two-valued function. The time dependent
signal of the sample after the column is measured. Subsequently, the correlation of
this signal and the modulation function is calculated. Of course, all other features
that are described above for the case of a method and an apparatus for determining
the ion mobility may be employed as well.
[0050] Other advantageous embodiments and combinations of features come out from the detailed
description below and the totality of the claims.
Brief description of the drawings
[0051] The drawings used to explain the embodiments show:
- Fig. 1 a, b
- a schematic view of an apparatus according to the invention and a block diagram showing
the steps of the method according to the invention, respectively,
- Fig. 2
- a correlation of a modulation function and an idealised signal from a single species
of ions,
- Fig. 3
- a schematic view of a linear feedback shift register that may be used to generate
a pseudorandom sequence for a modulation function,
- Fig. 4
- a sequence of maximum length and a corresponding idealised, filtered signal with enhanced
edges that is expected for an idealised signal of one ion species,
- Fig. 5
- correlations of a modulation function and an idealised signal from a single species
of ions, wherein for the different correlations the signal is sharpened with a different
sharpening parameter,
- Fig. 6
- a comparison of two correlations calculated for a measurement of a Leucine/Isoleucine-mixture,
once based on a filtered signal and once based on a non-filtered signal,
- Fig. 7a, b, c, d
- four different systematic deviations of the modulated ion beam from an ideal shape,
- Fig. 8
- simulated correlations illustrating that tailing and depletion of ions may cause a
false peak in the correlation which is not originating from a particular species of
ions,
- Fig. 9
- simulated correlations illustrating that the position of false peaks in the correlation
may be shifted by using a different tap set for the linear feedback shift register,
- Fig. 10
- four different modulation functions that are generated with the same linear feedback
shift register and the same tap set but with different sets of initial values, and
- Fig. 11
- a block diagram of a method that considers several possible optimisation options.
[0052] In the figures, the same components are given the same reference symbols.
Preferred embodiments
[0053] Figure 1a shows a schematic view of an ion mobility spectrometer 1 according to the
invention. This ion mobility spectrometer 1 may be used to execute a method according
to the invention in order to determine the mobility of ions. Figure 1 b shows a block
diagram of this method, illustrating the individual steps of the method.
[0054] The ion mobility spectrometer 1 comprises an ion gate 2, a drifting region 3, a detector
4 and a calculation unit 5. The drifting region 3 is confined by a tube 10. The ion
gate 2 is arranged on an opposite end of the tube 10 than the detector 4. The ion
gate 2 is of a known type. It comprises a grid of wires. If a voltage with opposite
signs is applied to neighbouring wires of the grid, ions of an ion beam 6 are prevented
of entering the tube 10. If there is no voltage applied to the wires of the grid,
the ions of the ion beam 6 may enter the tube 10. The switching of the ion gate 2
is controlled by a controller 7. The ion gate 2 may be switched between an open state,
where ions may pass the ion gate 2 and a closed state, where ions are prevented of
passing the ion gate 2. Those ions of the ion beam 6 that pass the ion gate 2 enter
the tube 10 and drift through the drifting region 3 to the detector 4 which generates
an ion signal. This ion signal is then passed to the calculation unit 5 for further
processing.
[0055] When performing a measurement, the ion gate 2 is controlled by the controller 7 to
switch according to a modulation function. This modulation function is a binary function
that may be represented as a sequence of bits having a value "1" or "0". A value "1"
corresponds to the open state of the ion gate 2, while a "0" corresponds to the closed
state of the ion gate 2. The modulation function is chosen such that its autocorrelation
is a two-valued function that has a peak at zero and otherwise a constant value. The
ion beam 6 approaches the ion gate 2 as a continuous ion beam. When entering the tube
10, it is modulated by the ion gate 2 to yield a modulated ion beam. In flight direction
of the ions, this modulated ion beam has a shape that corresponds to the modulation
function. The ions of the modulated ion beam are guided through the drifting region
3 and reach the detector 4, where a signal is generated. This signal is passed to
the calculation unit 5, where a correlation of the signal and the modulation function
is calculated. This correlation corresponds to the ion mobility spectrum.
[0056] As the autocorrelation of the modulation function is a two-valued function, the calculation
of the correlation of the signal and the modulation function does not introduce additional
features into the ion mobility spectrum. If, for example, the ion beam 6 comprises
one single species of ions, all ions take the same time for passing the drifting region
3. Accordingly, in an ideal measurement, where the modulated ion beam has exactly
the shape of the modulated function, the calculated correlation is a two-valued function
like the autocorrelation of the modulation function. But in contrast to the autocorrelation,
in the calculated correlation the peak position indicates the ions' time of flight
(see Figure 2).
[0057] As mentioned above, the modulation function is a binary function. More precisely,
it is a pseudorandom sequence of bits. It is generated by a linear feedback shift
register (LFSR) 30 which is incorporated in the controller 7. Figure 3 shows a schematic
representation of this LFSR 30. In the described embodiment, the LFSR 30 is a Fibonacci
implementation of an LFSR provided by a separate physical electronic circuitry. Alternatively,
it may be a Galois implementation. In a variant, it may be provided by some software
that is running on a computer instead of being provided by a separate physical electronic
circuitry. In other embodiments of the ion mobility spectrometer 1 an LFSR 30 may
be employed as well, but the modulation function generated by the LFSR 30 could for
example be a GMW sequence, a Welch-Gong transformation sequence, a Quadratic residue
sequence, a Sextic residue sequence, a Twin prime sequence, a Kasami power function
sequence, a Hyperoval sequence or a sequence derived from 3 or 5 maximum length sequences.
In the latter case for example, the sequence may be obtained by adding up the content
of corresponding bits of the 3 or 5 maximum length sequences. In that case, the addition
of two 1s or of two 0s may results in a 0, while the addition of a 0 and a 1 or of
a 1 and a 0 may result in a 1 (bitwise NAND). In order to achieve this addition, the
controller 7 may include an addition unit which is arranged after the LFSR 30.
[0058] Alternatively, the ion mobility spectrometer 1 shown in Figure 1 may comprise a store
for storing a predefined modulation function. In that case, the modulation function
may be generated by the LFSR 30 and stored in the store. When required, the modulation
function may be retrieved from the store. In a variant, the ion mobility spectrometer
may only comprise a store for storing a predefined modulation function and not comprise
the LFSR 30. Then, the modulation function may be generated by a separate LFSR like
the one shown in Figure 3. Subsequently, the modulation function may be permanently
stored in the store of the ion mobility spectrometer 1 as a predefined modulation
function. For a measurement, this predefined modulation function may be retrieved
from the store.
[0059] In a variant, another means than the above described LFSR 30 could be employed for
generating the modulation function. In such an embodiment, the same types of modulation
function could be used and the modulation function could be stored as described above.
[0060] As shown in Figure 3, the LFSR 30 has a number of bits 20.1, ... 20.5 which are connected
in series. Furthermore, the bits 20.1, ... 20.5 are connected by connections 22.1,
... 22.5 with XOR-functions 21.1, ... 21.5 that are themselves connected in series.
The connections 22.1, ... 22.5 can be individually switched on or off. Accordingly,
different connection patterns between the bits 20.1, ... 20.5 of the LFSR 30 and the
XOR-functions 21.1, ... 21.5 can be achieved by switching on or off the connections
22.1, ... 22.5. Each such connection pattern is called a tap set of the LFSR. For
generating a pseudorandom sequence, a particular tap set is chosen and the bits 20.1,
... 20.5 of the LFSR 30 are set to a set of initial values. Subsequently, based on
the values of the bits 20.1, ... 20.5 and based on the tap set, a bit-value is generated
by the XOR-functions 21.1, ... 21.5. This bit-value is fed to a first bit 20.5 of
the LFSR 30, while the values of the other bits 20.1, ... 20.4 of the LFSR 30 are
shifted by one bit towards the end of the LFSR 30. The last bit 20.1 of the LFSR 30
represents a bit of the pseudorandom sequence. By repeating the generation of a bit-value
from the current values of the bits 20.1, ... 20.5 and the tap set and by feeding
the generated bit-value to the LFSR 30, the pseudorandom sequence is generated.
[0061] In the described embodiment, the pseudorandom sequence generated by the LFSR 30 is
a sequence of maximum length. Accordingly, it has a length of 2
m -1 bits, where
m is the number of bits of the LFSR 30. For example, if
m = 7, the following tap sets are possible for obtaining a sequence of maximum length:
tap setm-7 1: [7, 6]
tap setm-7 2: [7, 4]
tap setm-7 3: [7, 6, 5, 4]
tap setm-7 4: [7, 6, 5, 2]
tap setm-7 5: [7, 6, 4, 2]
tap setm-7 6: [7, 6, 4, 1]
tap setm-7 7: [7, 5, 4, 3]
tap setm-7 8: [7, 6, 5, 4, 3, 2]
tap setm-7 9: [7, 6, 5, 4, 2, 1]
[0062] The numbers in these tap sets identify the open connections 22.1, ... 22.5 of the
bits 20.1, ... 20.5 with the XOR-functions 21.1, ... 21.5. In the given example, where
m = 7, the number 7 identifies the connection to the first bit 20.5 where the generated
bit-value is fed to (arrow), while the number 1 identifies the connection of the second
last bit 20.2 with the XOR-function 21.1. As shown in Figure 3, the output of the
LSFR 30 is always connected to the XOR-function 21.1 while the generated bit-value
is always fed to the first bit 20.5.
[0063] For generating a sequence, a set of initial values is chosen and the bits 20.1, ...
20.5 of the LFSR 30 are set accordingly. In this document, the sets of initial values
are denoted in the form of a decimal number. In order to set the bits 20.1, ... 20.5
of the LFSR 30, this number is to be represented in the form of a binary number.
[0064] In order to increase the resolution of the ion mobility spectrometer, the signal
can be filtered with a filter for enhancing the edges before the correlation between
the modulation function and the signal is calculated. The ion mobility spectrometer
1 shown in Figure 1 may therefore comprise a filter. This filter may be an n-element
finite difference filter, an edge enhancement filter or a filter using a different
type of sharpening algorithm. It may be incorporated in the calculation unit 5 or
may be a separate unit that is located between the detector 4 and the calculation
unit 5.
[0065] Figures 4, 5 and 6 illustrate the behaviour of the filter on the example of an n-element
finite difference filter. Figure 4 shows a sequence of maximum length (dashed line)
that is generated by the LFSR 30 shown in Figure 3 having a length of
m = 7. The continuous line shows a filtered signal that is expected for a perfect measurement
of one ion species being modulated with the shown sequence of maximum length. In reality,
the ions would reach the detector 4 with a delay that corresponds to the ion drift
times. Here in Figure 4, the filtered signal is shifted in time to correspond to the
sequence of maximum length in order to enable a comparison between the filtered signal
and the sequence of maximum length.
[0066] Since in the ion mobility spectrometer 1 the signal is measured in bins having a
specific width in time, the n-element finite difference filter comprises an algorithm
of the form
where
n is a measure for the width of the filter,
Di is the size of the signal's /
th bin and F
i is the filter-value's /
th bin. In order to obtain the filtered signal, each filter-value
Fi is added to the corresponding bin
Di of the measured signal. When doing so, the filter-values
Fi and the signal D
i are multiplied with a weight factor before adding the filter-values to the signal.
These weight factors are based on
n, the width of the filter, with 0 <=
n<=
nmax:
[0067] Figure 5 shows calculated correlations of the modulation function and the signal
shown in Figure 4 with the signal being filtered with different sharpening parameters
n. The peak indicating the time of flight of the ions becomes sharper with increasing
sharpening parameter
n. But at the same time, there is an overshoot 40.1, 40.2 on both sides of the peak
which becomes stronger with increasing sharpening parameter
n. Therefore, the filtering has the effect that peaks originating from ions having a
similar time of flight may be resolved better. This is illustrated in Figure 6 on
the example of an ion mobility spectrum for a Leucine/Isoleucine-mixture, where the
peaks that represent the time of flight of Leucine and Isoleucine, respectively, can
be resolved better if the signal is filtered.
[0068] In a real measurement, the modulated ion beam has never the perfect shape of the
modulation function. There will always be some systematic deviations from the perfect
shape. Four types of such deviations are illustrated in Figures 7a, 7b, 7c and 7d.
In Figure 7a, a deviation is shown which is caused by depletion. In this case, when
the ion gate is switched into the open state, it takes some time before ions start
to enter the drifting region. Accordingly, the modulation function's bits in the modulated
ion beam get a slope towards lower times of flight. As another possible systematic
deviation, Figure 7b shows a modulated ion beam that is distorted by a delayed response
of the ions. This may occur due to a non-uniform gas flow in the drifting region or
due to other reasons. It distorts the modulation function's bits in the modulated
ion beam in a manner similar to a rectangular signal being distorted by an RC filter.
A further type of deviation is a tailing of the ions. As shown in Figure 7c, in this
case, some ions get delayed when passing the drifting region. Therefore, the modulation
function's bits in the modulated ion beam obtain a tail towards higher times of flight.
A fourth type of systematic deviations is caused by diffusion of the ions. Figure
7d illustrates how in that case the edges of the modulation function's bits in the
modulated ion beam become diffused during the ions' passage through the drifting region.
[0069] From these four types of systematic deviations, the diffusion is the only one which
is symmetric in time. Accordingly, it causes only a broadening of the peaks in the
calculated correlation. This broadening may be at least partially taken into account
for by filtering the signal before calculating the correlation. The other three types
of systematic deviations may as well cause a broadening of the peaks which may be
taken into account for by filtering the signal and thus sharpening the correlation.
But additionally, due to their asymmetry in time, they cause a shifting of the peak
positions and may cause features at other positions of the correlation. For example,
as shown in Figure 8, tailing and depletion may cause a false peak 50.1, 50.2 in the
correlation that is not originating from a particular species of ion. Additionally,
both these deviations may cause false features 51.1, 51.2 in the baseline of the correlation.
In order to take into account for the shifting of the peaks, the false peaks 50.1,
50.2 and the false features 51.1, 51.2, there are different approaches to be chosen.
The shifting for example may be taken into account for by calibrating the ion mobility
spectrometer accordingly.
[0070] Figure 9 illustrates an approach for how to deal with a false peak 50.1, 50.3, ...
50.6. It shows simulated correlations that are calculated by assuming a measurement
of a single species of ions, wherein some of the ions are tailing. These simulated
correlations are based on modulation functions that are pseudorandom sequences of
maximum length. The sequences are generated by an LFSR 30 as shown in Figure 3. The
LSFR 30 has a length of 7 bits. The difference between the simulated correlations
is that for each correlation, a different tap set of the LFSR 30 is used for generating
the pseudorandom sequences of maximum length. As shown, the position of the false
peak 50.1, 50.3, ... 50.6 depends on the tap set of the LFSR 30. Since the position
does not depend on the set of initial values used for generating the pseudorandom
sequences of maximum length, it is sufficient to choose a tap set such that the false
peak 50.1, 50.3, ... 50.6 is located outside of an interval of interest. In Figure
9, if the interval of interest is for example between a drift time of 400 and 800
arbitrary units, the tap sets [7, 4], [7, 6, 4, 2] or [7, 6, 5, 4] may be used because
the position of the false peak 50.1, 50.3, 50.6 is then located outside of the interval
of interest.
[0071] One approach for how to deal with false features 51.1, 51.2 like the ones shown in
Figure 8 is to choose a tap set of the LFSR 30 such that the false features 51.1,
51.2 have a minimal height. Another approach which may additionally be employed is
illustrated in Figure 10, where four different modulation functions are shown. All
four modulation functions are pseudorandom sequences of maximum length that have been
generated with the LFSR 30 as shown in Figure 3. The LFSR 30 has had a length of 7
bits and the tap set [7, 6, 4, 1] has been used. For each of the four modulation functions
shown in Figure 10, a different set of initial values has been used. As a consequence,
the average of the obtained modulation functions provides fewer steps than the individual
modulation functions. This effect can be used in the method for obtaining an ion mobility
spectrum. When doing so, a measurement is repeated in cycles by using for each cycle
a different modulation function that is generated by using a different set of initial
values. Subsequently, the obtained correlations are added to a total correlation.
Since for each modulation function, false features 51.1, 51.2 like the ones shown
in Figure 8 are located at different positions of the baseline of the calculated correlation,
the false features 51.1, 51.2 get averaged out.
[0072] In order to implement this averaging option into an ion mobility spectrometer, the
latter may comprise a summation unit for calculating the total correlation from the
correlations obtained from the measurements with different modulation functions. This
summation unit may be incorporated into the calculation unit 5 (see Figure 1 a) or
it may be a separate unit arranged after the calculation unit 5.
[0073] When considering these optimisation options, the method according to the invention
which is shown in Figure 1 b may be extended. Figure 11 shows a scheme of a method
that considers these options. The individual steps of the method are illustrated.
[0074] In this extended method, an LFSR is used for generating the modulation function.
Accordingly, the tap set of the LFSR is chosen first. This choice is based on the
criterions that any false peak caused by tailing or depletion of the ions is located
outside of the interval of interest of the correlation and that false features caused
by tailing, depletion or a delayed response of the ions have a low intensity in the
correlation. In a second step, different sets of initial values of the LFSR are chosen.
These sets are chosen such that false features caused by tailing, depletion or a delayed
response of the ions are located at different positions in the correlation. Since
the false peaks and the false features depend on systematic deviations of the modulated
ion beam from a prefect shape, they may be simulated according to the characteristics
of the ion mobility spectrometer that is used. Accordingly, the choice of the tap
set of the LFSR and of the sets of initial values may be based on such simulations.
[0075] Once the tap set of the LFSR and the sets of initial values are chosen, some steps
of the method are repeated in cycles. During each cycle, a modulation function is
generated first. This modulation function is based on the preliminary chosen tap set
and on one of the preliminary chosen sets of initial values. During each cycle, the
set of initial values is different. Once the modulation function is generated, the
ion beam is modulated by the ion gate according to the modulation function. The modulated
ion beam is then guided through the drifting region and a signal of the ions is measured
after the ions have passed the drifting region. Subsequently, the measured signal
is sharpened and the correlation of the modulation function and the sharpened signal
is calculated. In each cycle, the correlation is either stored in a separate store
or fed directly to a summation unit for adding the correlations calculated during
the cycles. If during each cycle, the correlation is stored in a separate store, the
correlations may be fed to the summation unit after the last cycle is executed. Finally,
all correlations obtained during the cycles are added by the summation unit. The resulting
total correlation corresponds to the ion mobility spectrum.
[0076] In this extended method, the step of generating the modulation functions may be executed
before the measurements are repeated in cycles. In that case, the modulation functions
are stored in a store before repeating the measurement in cycles. Subsequently, during
each cycle, a different modulation function is retrieved from the store.
[0077] In a further embodiment of the above described ion mobility spectrometer, the detector
is a mass spectrometer. This enables to obtain an ion mobility spectrum and a mass
spectrum of the same ions. The mass spectrometer employed may be a time-of-flight
mass spectrometer, a quadrupole mass spectrometer, an ion trap mass spectrometer or
another type of mass spectrometer. In order to optimise the performance of the ion
mobility spectrometer and the mass spectrometer, the mass spectrometer is capable
of obtaining mass spectra with a high repetition rate. In particular, it may be permanently
operatable with this high repetition rate or it may be operatable with this high repetition
rate for at least the time interval that is required for measuring one ion mobility
spectrum by using the entire modulation function. For example, the modulation function
of the ion mobility spectrometer may comprise bits with a length of about 250 µm.
In this case, the mass spectrometer may repeatedly obtain a mass spectrum within 250
µm or within a fraction of 250 µm. The latter case is particularly advantageous, if
the time-resolution of the obtained ion mobility spectra is better than 250 µm. For
example, if the ion mobility spectra have a time-resolution of 50 µm caused by diffusion
of the ions in the drifting region, the mass spectrometer may obtain mass spectra
with a repetition rate of 50 µm or a fraction thereof. Of course, these particular
bit length, time-resolution and repetition rates are only examples for illustration
purposes. They may be adapted to the particular requirements of the measurements to
be performed.
[0078] In summary, it is to be noted a method and an apparatus are provided that allow for
determining an ion mobility with a higher signal to noise ratio while providing the
same measurement speed as known from the prior art.
1. A method for determining a mobility of ions, including the steps of:
a. modulating an ion beam (6) with an ion gate (2) which is controlled by a modulation
function for generating a modulated ion beam,
b. guiding said modulated ion beam through a drifting region (3),
c. measuring a signal of said modulated ion beam after said modulated ion beam has
passed said drifting region (3),
d. calculating a correlation of said modulation function and said signal in order
to
determine said mobility of said ions,
characterised in that an autocorrelation of said modulation function is a two-valued function.
2. The method according to claim 1, characterised in that said modulation function is a pseudorandom sequence.
3. The method according to claim 2, characterised in that said modulation function is a maximum length sequence, a GMW sequence, a Welch-Gong
transformation sequence, a Quadratic residue sequence, a Sextic residue sequence,
a Twin prime sequence, a Kasami power function sequence, a Hyperoval sequence or a
sequence derived from 3 or 5 maximum length sequences.
4. The method according to one of claims 1 to 3, characterised by a step of enhancing edges of said signal with a filter by filtering said signal before
calculating said correlation.
5. The method according to claim 4, characterised in that said filter is an n-element finite difference filter, edge enhancement filter, or
a filter using a different type of sharpening algorithm.
6. The method according to one of claims 1 to 5, characterised in that an interval of interest of possible ion drift times is chosen from said correlation.
7. The method according to claim 6, characterised by the step of selecting said modulation function such that as many as possible false
peaks (50.1, 50.2, 50.3, 50.4, 50.5, 50.6) in said correlation are located outside
of said interval of interest.
8. The method according to one of claims 1 to 7, characterised by a step of selecting said modulation function such that false features (51.1, 51.2)
in said correlation have a low height.
9. The method according to one of claims 1 to 8,
characterised in
a. repeating said steps in cycles, wherein during each cycle, said ion beam (6) is
modulated with said ion gate (2) being controlled by a different modulation function
from a set of modulation functions for generating a different modulated ion beam,
and in
b. adding said correlation which is calculated during each said cycle to a total correlation
in order to determine said mobility of said ions.
10. The method according to claim 9, characterised in performing a preliminary step before repeating said cycles, wherein said set of modulation
functions is selected such that for each modulation function, false features (51.1,
51.2) in said correlation are located at different positions of said correlation and
thus said false features (51.1, 51.2) are averaged out in said total correlation.
11. The method according to one of claims 1 to 10, characterised in that said correlation is calculated by calculating a circular cross correlation, an inverse
Hadamard-transformation a Fourier transformation, a Laplace transformation or an M-transformation.
12. Apparatus (1) for determining a mobility of ions, including:
a. an ion gate (2) which is controlled by a modulation function for generating from
an ion beam (6) a modulated ion beam,
b. a drifting region (3) through which said modulated ion beam is guidable,
c. a detector (4) by which a signal of said modulated ion beam is measurable after
said modulated ion beam has passed said drifting region (3),
d. a calculation unit (5) by which a correlation of said modulation function and said
signal is calculable in order to determine said mobility of said ions,
characterised in that an autocorrelation of said modulation function is a two-valued function.
13. The apparatus (1) according to claim 12, characterised by a linear feedback shift register (30) by which a pseudorandom sequence is generatable
for the use as said modulation function.
14. The apparatus (1) according to claim 12 or 13, characterised in that before said correlation is calculable, a filter for enhancing edges of said signal
is applicable by said calculation unit (5) to said signal.
15. The apparatus (1) according to one of claims 12 to 14,
characterised by
a. a control unit (3) by which a repetition in cycles of steps is controllable, said
steps including generating said modulated ion beam with said ion gate (2), guiding
said modulated ion beam through said drifting region (3), measuring said signal with
said detector (4) and calculating said correlation of said modulation function and
said signal, and
b. a summation unit by which a total correlation is calculable in order to determine
said mobility of said ions, said total correlation being a sum of said correlations
calculated during said cycles.
16. The apparatus (1) according to one of claims 12 to 15, characterised in that said detector (4) is a mass spectrometer.
17. The apparatus (1) according to claim 16, characterised in that said mass spectrometer is a time-of-flight mass spectrometer.
18. The apparatus (1) according to claim 16 or 17, characterised in that the mass spectrometer allows for determining ion mass spectra with a repetition rate
that corresponds to a time resolution of the obtainable ion mobility spectra or to
a fraction thereof.